0405

Combined Cardiac CINE and T­1, T2, and M0 Mapping with MR Fingerprinting
Jesse Ian Hamilton1, Mark Griswold1,2, and Nicole Seiberlich1,2

1Biomedical Engineering, Case Western Reserve University, Cleveland, OH, United States, 2Radiology, University Hospitals, Cleveland, OH, United States

Synopsis

A method is introduced for combined cardiac CINE and T1, T2, and M0 quantification using MR Fingerprinting throughout the cardiac cycle. Data are acquired continuously using an MRF acquisition and binned by cardiac phase using an external ECG. Undersampled MRF images from all TRs are non-rigidly registered to a consistent cardiac phase before pattern matching in order to generate CINE T1, T2, and M0 maps. CINE-MRF has a breathhold duration (10s) and temporal resolution (25 cardiac phases) similar to conventional CINE scans, while providing both relaxation time measurements and functional information.

Introduction

In a typical CMR exam, multiple CINE scans covering the whole LV and diastolic T1 and T2 maps for a few slices are collected in separate acquisitions. Cardiac MR Fingerprinting (cMRF) has previously been introduced for simultaneous T1, T2, and M0 quantification.1 However, as with conventional mapping techniques, cMRF uses prospective ECG gating and only provides maps from one cardiac phase. Inspired by CMR multitasking,2 this study proposes a method termed CINE-MRF for continuous relaxation time mapping throughout the cardiac cycle, enabling the generation of T1 and T2 maps and functional images during one breathhold. CINE-MRF uses non-rigid registration to improve SNR and reduce aliasing artifacts in the motion-resolved maps.

Methods

CINE-MRF data are collected with a time-varying FISP-based sequence similar to prior cMRF work3 except that no trigger delays are used. A total of 1920 TRs are collected in 10s using a spiral trajectory4 with golden angle rotation.5 Figure 1 shows key steps in the reconstruction. First, k-space data are binned by cardiac phase using an external ECG signal. Second, initial maps for each phase are estimated using a low rank reconstruction6 with total variation regularization along the cardiac motion dimension. Although these maps have artifacts, they are sufficient for visualizing cardiac motion. Third, the maps are combined to generate a CINE movie, and displacement fields are calculated via non-rigid registration (NiftyReg with a mutual information objective function7) between every pair of cardiac phases. Fourth, k-space data are gridded using the NUFFT.8 Undersampled images from all TRs are registered using the displacement fields to a consistent cardiac phase before dot product matching with the dictionary.9 Registration and matching are repeated for each cardiac phase to generate CINE T1, T2, and M0 maps.

Eleven volunteers were scanned at 3T (Siemens Skyra) after obtaining written informed consent in a HIPAA-compliant, IRB-approved study. Scans were acquired from one short-axis slice (300mm2 FoV, 1.6x1.6x8.0mm3). CINE-MRF data were reconstructed to yield 25 cardiac phases, similar to a conventional CINE. ECG-gated cMRF data were also collected with a 255ms acquisition window and 16-heartbeat breathhold, along with conventional T1 and T2 maps using MOLLI and T2-prepared FLASH (Siemens MyoMaps).10,11 These ECG-gated scans were collected with scan windows placed in both diastole and systole. Finally, a conventional CINE was acquired with 25 cardiac phases, bSSFP readout, 45-60 flip angle, and TR/TE 2.9/1.4ms. Relaxation times were measured within a septal region of the myocardium, and single slice ejection fractions (EF) were compared between CINE-MRF and standard CINE datasets.

Results

Figure 2 shows representative in vivo maps in diastole and systole from each technique. Figure 3 displays CINE-MRF maps over several cardiac phases. Average relaxation times measured in diastole and systole are presented in Figure 4. T1 was slightly higher with CINE-MRF than ECG-gated cMRF (both higher than MOLLI). T2 values were slightly lower with CINE-MRF than ECG-gated cMRF (both lower than T2-prepared FLASH). Paired t-tests indicated no significant differences between diastolic/systolic T1 measured with any technique. No significant differences were observed between diastolic/systolic T2 values using CINE-MRF. However, the average T2 was 1.2ms higher in diastole vs. systole using ECG-gated cMRF and 2.3ms lower in diastole vs. systole using T2-prepared FLASH, and these differences were significant. Figure 5 summarizes the EF measurements. A Bland-Altman analysis12 found good agreement between both standard CINE and CINE-MRF, with bias 1.3% (CINE-MRF EF larger than standard CINE) and 95% limits of agreement (-4.0%,6.7%).

Discussion

This study demonstrates the feasibility of simultaneous T1, T2, M0, and functional quantification throughout the cardiac cycle using MRF. The temporal resolution of CINE-MRF (40ms/frame assuming a 60bpm heart rate) and 10s breathhold are comparable to clinical CINE scans. EF measurements agreed well with conventional CINE. Relaxation times from CINE-MRF were consistent among volunteers. However, T1 was systematically higher and T2 systematically lower than ECG-gated cMRF. These effects could arise from through-plane motion, which was shown to produce erroneously low T2 values in the brain with FISP-MRF13, and could be addressed by a 3D acquisition. CINE-MRF could potentially provide insight into T1 and T2 changes during the cardiac cycle. Future work will investigate how non-rigid registration affects the ability to measure dynamic T1 and T2 changes. While ECG-gated cMRF generates a new dictionary for every scan that incorporates the subject’s cardiac rhythm, CINE-MRF has fixed sequence timings and uses a precomputed dictionary.

Conclusion

A technique is introduced for joint CINE and relaxation time mapping over 25 cardiac phases in a 10s breathhold, which is comparable to clinical CINE scans. This method could enable whole-LV motion-resolved mapping and greatly simplify CMR exams.

Acknowledgements

NIH R01HL094557, R01DK098503,R01EB016728; NSF CBET 1553441, Siemens Healthineers (Erlangen, Germany)

References

  1. Hamilton JI, Jiang Y, Chen Y, et al. MR fingerprinting for rapid quantification of myocardial T1, T2, and proton spin density. Magn Reson Med. 2017;77:1446-1458.
  2. Christodoulou AG, Shaw JL, Nguyen C, et al. Magnetic resonance multitasking for motion-resolved quantitative cardiovascular imaging. Nat Biomed Eng. 2018;2:215-226.
  3. Jiang Y, Ma D, Seiberlich N, Gulani V, Griswold MA. MR fingerprinting using fast imaging with steady state precession (FISP) with spiral readout. Magn Reson Med. 2015;74(6):1621-1631.
  4. Hargreaves B. Variable-Density Spiral Design Functions. http://mrsrl.stanford.edu/~brian/vdspiral/. Published 2005. Accessed June 1, 2017.
  5. Winkelmann S, Schaeffter T, Koehler T, Eggers H, Doessel O. An optimal radial profile order based on the Golden Ratio for time-resolved MRI. IEEE Trans Med Imaging. 2007;26(1):68-76.
  6. Hamilton JI, Jiang Y, Ma D, et al. Low Rank Compressed Sensing Reconstruction for More Precise Cardiac MRF Measurements. Proc. ISMRM 2017; #554.
  7. Modat M, Ridgway GR, Taylor ZA, et al. Fast free-form deformation using graphics processing units. Comput Methods Programs Biomed. 2010;98(3):278-284.
  8. Fessler J, Sutton B. Nonuniform fast Fourier transforms using min-max interpolation. IEEE Trans Signal Process. 2003;51(2):560-574.
  9. Ma D, Gulani V, Seiberlich N, et al. Magnetic resonance fingerprinting. Nature. 2013;495(7440):187-192.
  10. Messroghli DR, Radjenovic A, Kozerke S, Higgins DM, Sivananthan MU, Ridgway JP. Modified look-locker inversion recovery (MOLLI) for high-resolution T 1 mapping of the heart. Magn Reson Med. 2004;52(1):141-146.
  11. Siemens Medical Solutions USA. MyoMaps. https://usa.healthcare.siemens.com/magnetic-resonance-imaging/options-and-upgrades/clinical-applications/myomaps. Published 2018. Accessed February 27, 2018.
  12. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1(8476):307-310.
  13. Cruz G, Jaubert O, Schneider T, Botnar RM, Prieto C. Rigid motion-corrected magnetic resonance fingerprinting. Magn Reson Med. September 2018.

Figures

Steps in the CINE-MRF reconstruction. (A) First, k-space data are sorted into 25 phases using an external ECG signal (only 3 phases are shown for clarity). (B) Next, a low rank reconstruction is performed using the SVD of the dictionary to generate initial estimates of the parameter maps. (C) A CINE movie is generated by combining the T1, T2, and M0 maps at each cardiac phase. Non-rigid displacement fields are calculated between every possible pair of phases. (D) Undersampled images from all TRs are registered to a consistent cardiac phase using the displacement fields and matched to the dictionary to obtain motion-resolved maps.

In vivo maps from a volunteer at 3T using MyoMaps (MOLLI for T1 mapping and T2-prepared FLASH for T2 mapping), ECG-gated cardiac MRF, and CINE-MRF without prospective ECG gating. Maps are shown in (A) diastolic and (B) systolic cardiac phases.

T1, T2, and M0 maps acquired from a volunteer using CINE-MRF are shown over the cardiac cycle. For this scan, the average RR interval was 1120+/- 45ms. CINE-MRF maps were reconstructed with 25 cardiac phases (temporal resolution 45ms/frame)

Ejection fractions from 11 volunteers measured from a single short-axis slice using a standard CINE scan and CINE-MRF. Standard CINE data for volunteers #10 and #11 are not available.

Myocardial wall (A) T1 and (B) T2 values averaged over all volunteers at 3T using MyoMaps (i.e. MOLLI for T1 mapping, and T2-prepared FLASH for T2 mapping), ECG-gated cMRF, and CINE-MRF without prospective gating. The errors bars denote the standard deviation of the mean relaxation times from each volunteer. An asterisk denotes a significant difference (p < 0.05) between diastolic and systolic relaxation times according to a paired t-test.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)
0405